Quentin Stafford-Fraser was a researcher in the Computer Laboratory at the University of Cambridge in 1991, working on the top floor of the old Corn Exchange Street building, when he wrote a short program called XCoffee to spare himself and his colleagues the walk down to the Trojan Room. The coffee pot lived one floor below. The pot was small, the department was large, and by the time anyone in a distant office got there, the jug was often empty and someone else was already brewing a fresh batch. Stafford-Fraser and his colleague Paul Jardetzky rigged a grey-scale video camera, wired it into a frame-grabber card on a spare Acorn computer, and pushed a fresh image to any workstation on the local network that asked for one.
It was, by the count of the University of Cambridge itself, the first webcam. And it was pointed at a coffee jug.
The stairwell that started it
The Trojan Room was a shared coffee area on a lower floor of the Computer Lab. Researchers on other floors — some of them working on the earliest days of the ATM network, some on distributed computing, some on what would later be called ubiquitous computing — had a recurring, deeply unglamorous problem. You would put down your work, walk down the stairs, and find the pot drained. The frustration of a wasted trip is the kind of small, repeating annoyance that the brain can treat as a genuine obstacle to a goal. Repeated often enough, that annoyance produces resignation, or a workaround.
Stafford-Fraser and Jardetzky chose the workaround.
The camera they used was a spare monochrome video unit lying around the lab. The frame-grabber was slotted into an Acorn Archimedes. The software Stafford-Fraser wrote — XCoffee, on the server side, and XCoffee-client on the desktop side — captured a fresh frame from the pot and served it to anyone on the department network who wanted to check the level before setting off down the stairs.
The image was tiny. A postage stamp, in effect: low resolution, in shades of grey, with the jug in the centre of the frame and, if you squinted, a rough sense of how much dark liquid remained.
A local hack that leaked out
For three years, the coffee pot camera was an internal joke, useful only to people already inside the Cambridge Computer Lab. Then, in 1993, two other researchers at the lab — Daniel Gordon and Martyn Johnson — modified the setup to serve the image over HTTP, the protocol Tim Berners-Lee had introduced at CERN a few years earlier. Once the pot was on the open web, anyone with a browser and a network connection could look at it.
People did. In enormous numbers.
Word spread through Usenet groups and early web directories. The Trojan Room coffee pot page, hosted by the Computer Lab, was viewed by millions over its lifetime, from researchers in Tokyo to journalists in New York. The German magazine Der Spiegel ran a piece on it. The BBC covered it. Visitors to Cambridge asked to be shown the actual jug. The camera and the pot became, briefly, one of the most looked-at objects in the world — and the object in question was a stained filter-coffee machine that produced, by most accounts, rather mediocre coffee.
Why a tiny grey image mattered
The mechanism behind the fame is worth stopping on. The web in 1993 was almost entirely text. Marc Andreessen’s Mosaic browser had only recently made inline images practical. A live, updating photograph of a real object — even a boring one — was something new. It was the first time the network became a window rather than a page.
The coffee pot was not sending a stream in the modern sense. Each visit to the page fetched a fresh still image, refreshed at regular intervals. There was no video, no sound, no streaming protocol worth the name. And yet the psychological effect on early viewers was profound: for the first time, a computer screen showed something happening in a specific place, right now, thousands of miles away. Someone in California could see, with a delay of seconds, whether a jug of coffee in a corridor in England was full or empty.
That is the trick every webcam since has inherited. The Cambridge pot proved that a network could carry presence.
The problem-solving pattern behind it
The invention of the webcam was not a research project. There was no grant, no proposal, no roadmap. It was, in the language of problem-solving strategies, a heuristic solution — a rule-of-thumb workaround built by people with the tools already on their desks, applied to a repeating irritation. Stafford-Fraser and Jardetzky did not sit down to invent a new medium. They wrote a small program to save themselves a walk.
That pattern — small, personal, local frustration producing infrastructure that eventually reshapes the world — runs through much of the early internet. The department needed to share files, so someone wrote FTP. The physicists at CERN needed to link documents, so Berners-Lee wrote HTTP. The Cambridge researchers needed to know whether it was worth walking down the stairs, so two of them pointed a camera at a jug. Building on intuitive problem-solving, insight often arrives during moments of stalled attention: on a walk, in the shower, waiting for something else to finish. A wait for coffee is, in that sense, a small laboratory for exactly the kind of thinking that produces workarounds.
Silicon Canals has covered the deeper backstory of the Cambridge coffee cam before, and the mundane origin is the point. The story is not that the researchers foresaw the modern webcam economy. They did not. They solved a nuisance and left the artefact running.
What the camera saw for eleven years
The Trojan Room coffee pot camera ran continuously from 1991 until August 2001. It was switched off when the Computer Laboratory moved out of the old Corn Exchange Street building and into the new William Gates Building on the West Cambridge site. The final image, captured that morning, showed the pot from the same angle it had held for a decade. Someone, on the last day, wrote a script that let visitors watch the machine’s final hours.
The pot itself was auctioned. Der Spiegel bought it and had it refurbished for its Hamburg offices. The Krups machine — a German-made drip filter, of the kind sold in department stores across Europe — was made functional again, and reportedly served coffee to journalists for years afterwards.
The camera, the frame-grabber, and the software were retired. The URL still exists as an archive page. The pot’s descendants number in the hundreds of millions.
The line from a coffee jug to a billion video calls
By 2020, when the world shifted to remote work under pandemic conditions, the webcam had become the primary interface between colleagues, families, doctors and patients, teachers and students. Zoom’s daily meeting participants rose from 10 million in December 2019 to 300 million by April 2020. Every one of those calls carried the DNA of the Cambridge experiment: a camera pointed at a fixed spot, an image pushed over a network, a viewer at the other end deciding whether it was worth acting on what they saw.
The line is not metaphorical. The engineering choices Stafford-Fraser and Jardetzky made — capture, compress, serve on request — are, in stripped-down form, still the architecture of modern video streaming. What has changed is bandwidth, resolution, frame rate, and audience size, not the basic idea.
The idea itself came from stairs, a jug, and the entirely reasonable human wish not to make the trip for nothing.
The mundane engine of invention
There is a temptation, looking back, to make the Cambridge story bigger than it was. To say the researchers saw what was coming. They did not. Stafford-Fraser has said, in interviews over the years, that he found the fame of the coffee pot slightly embarrassing — a piece of undergraduate-level engineering that eclipsed the serious research going on around it. In a 2001 personal account written for the Computer Lab, he described XCoffee as a program he’d written in an afternoon.
That is the shape of most first inventions of a medium. Samuel Morse’s opening telegraph message in 1844 was not a manifesto but a Bible verse — a fact Silicon Canals has traced in its history of the first telegraph transmission. The first email, the first text message, the first tweet: all were sent to prove the system worked, not to say anything. The Cambridge coffee pot fits the pattern exactly. The camera was there to show that the pot was there. That was the whole content.
What the pot left behind
The Krups coffee machine is now in a private collection, having passed through Der Spiegel‘s ownership. The Computer Laboratory maintains an archive page describing the experiment. Quentin Stafford-Fraser went on to co-found several companies and to work on video conferencing systems, which is, in a sense, the direct professional descendant of what he built in 1991. Paul Jardetzky moved into other computing work. Neither man became rich from the invention. They did not patent it. They did not, in the moment, think of it as an invention at all.
The next time you join a video call from a kitchen table, or check a security camera on your phone, or watch a livestream of a nesting osprey a continent away, the technical lineage runs back through millions of iterations to a grey-scale image of a stained glass jug, refreshed at regular intervals, on a corridor in Cambridge, watched at first by perhaps a dozen people who simply wanted to know whether it was worth going downstairs.
The pot was usually about half full.
